Testimony before the Committee on Science, Space and Technology Subcommittee on Technology, Environment and Aviation U.S. House of Representatives
Hearing on Communications and Computer Surveillance, Privacy and Security
May 3, 1994
David J. Farber The Alfred Fitler Moore Professor of Telecommunication Systems University of Pennsylvania 200 S. 33 rd Street Philadelphia PA 19104-6389 Net: [email protected]
Good morning Mr. Chairman and members of the subcommittee. I want to thank the Committee for inviting me to testify today.
I should start off by stating that the views I will give are my own, and not necessarily those of any organizations that I am affiliated with -- although I hope they agree with my views. I am speaking merely as a professional with long experience in the computer and communications field.
My background has been focused on the understanding of and the development of technology in the computer and communications area. I started my career in 1956 with a BS in General Engineering from Stevens Institute of Technology. Due to the influence of a friend, I interviewed Bell Telephone Laboratories even though I was accepted into graduate school. I accepted a very challenging job at Bell Labs helping design the worlds first electronic switching system which was later installed at Morris, Illinois. It was the ENIAC of the computer-based telephone systems we have today. I later went on to do pioneer work in programming languages and after a period at the RAND Corp. I left industry to join the University of California at Irvine (UCI) as a faculty member. At UCI I was responsible for the conceptualization of the first operational distributed computer system sponsored by the NSF -- the DCS system which was the first use of client-server ideas, along with the first micro kernel and the first ring-type Local Area Network. On my journey to the University of Pennsylvania where I hold the Chair of The Alfred Fitler Moore Professor of Telecommunication Systems, I collaborated in the creation and operation of CSNET and the NREN as well as co-authoring the proposal for the Gigabit Testbeds. I have served on more industrial and government advisory committees than is reasonable and am a Fellow of the IEEE. I serve on the Board of Trustees of both the Electronic Frontier Foundation and the Internet Society.
My career has been focused on the understanding of, and the development of, technology in the communications area. I have co-founded several of the data networks which have led to the National Information Infrastructure and have taken part in the creation of some of the technology that makes it possible. For the last several years I have also been increasingly involved with the policy problems and opportunities that this technology has created. So while my remarks will focus largely on the technical aspects of the Clipper and the Digital Telephony initiatives I would like to start out by expressing some of my views on policy and also my views as a concerned citizen.
II. Public Policy Challenges Posed By New Information & Communication Technologies
We are at a critical stage in the evolution of a capability that could rival the industrial revolution in bringing both good and bad to the citizens of this nation. The industrial revolution brought unheard of prosperity to parts of the world and pain and suffering also. The information revolution, and it is a revolution, can bring a burst of prosperity and vigor to a world which is increasingly stuck in a morass of slow growth, regionalization and trade conflicts.
The information revolution is a stew based on a blend made possible by increasingly complex communications technology mixed with increasingly more powerful computers. The National Information Infrastructure (NII) and it's sister the Global Information Infrastructure (GII) are among the exciting developments arising out of this technology "stew". All in all, we have an unprecedented transfer of power into the hands of the public and the government that results from these technologies. Computer power that served the entire technical staff of Bell Labs when I was there is now available in my hand. Communications capabilities only dreamed of in the Saturday movie serials is now readily available worldwide. This vastly increased power in the hands of the public has created difficulties for the policy makers. While technology increases in power at a rate of two times per year, our ability to understand the policy implications of this, sadly, grows very slowly.
This has shown itself recently in the debate about the Clipper encryption initiative and the Digital Telephony proposals both put forward by the Clinton/Gore administration and backed heavily by the law enforcement community. A careful look at each of these initiatives illuminates the fundamental national issues that are being fought out.
It is critical that this discussion be brought out into that part of the policymaking process which has always represented the citizens -- the Congress. I take as one of my guidelines in thinking about these issues, a quote by Ben Franklin -- the Founder of the University I am at -- "They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety."
A. Clipper and the proposed escrow system
The proposed key escrow system that is part of the "Clipper" proposal for providing securable telephony has been subject to many and often loud debates in the press and in the Internet world. Many worry that such an escrow system could be vulnerable to misuse by a future administration or overzealous law enforcement organizations. The result of any such misuse would be a major blow to the privacy rights of our citizens. The counter-argument is based on arguments that it is no worse than we now have -- namely no cryptographic protection in widespread use, and a fear of increasing difficulties due to technology and the ease of computer supported voice encryption. Clearly the issues are basic to the way we will live in the future that is, what will happen to both our liberty and our safety. Currently the discussants have taken extreme positions -- as is normal for the beginning of such a critical discussion. Neither side is capable of explaining to the other the reasons for their support of or concerns about the proposal.
Thus I believe that we must carefully examine the proposed escrow system to insure that we have not done the trade that Franklin warned us about, but rather, that we have used all available technology to ensure that we end up with more safety and more liberty.
The Administration's Clipper proposal to create an escrowed cryptographic system for telephony came out of the blue, at least as far as my community is concerned. When I sent out the first announcement of the Clipper proposal to my "interesting people" mailing list (a list that goes widely to senior people in academia, government and industry) my only comment was "basically the issue is 'do you trust your government'". One of the replies I received from a 23 year old person I know well said that "to my generation, the term 'trust government' has no meaning". Sad but true!!.
The Clipper announcement was sketchy, lacking substance and real form. We all "waited for the beef" but there was unfortunately no such beef. The main ingredient of the initiative and it's most controversial aspect -- the escrow system -- was not yet really designed. The reaction was, to put it mildly, mostly shock and outrage and even sadness. There were insufficient details about the Clipper system to enable us to make a competent technical evaluation of the overall security of the proposal, including all of its escrow management procedures. Those of us in industry and in academe, see our role as helping to assess for both the corporate community and individual users, the security and privacy of computer and communications systems. To this date, we have been unable to perform this function because most of the details of Clipper are shrouded in secrecy.
B. Where are we and where can we go
One of the facts of life in the Information Age is that powerful tools such as encryption are and will be increasingly more available to law-abiding citizens, business users, people all over the world, and, of course, to criminals. This does pose challenges to law enforcement, but there's nothing we can do to stop this technology. All that the government can do is to slow things down. We have to face it. Perhaps we can slow it down and leave it to our children to face these facts of life. But in my view we have created the world that presents these challenges and we should attempt to solve the problems and to leave a better world to our children.
We should not, however, underestimate the depth of the problem that sophisticated encryption technology and the information age thrusts upon us. On the one hand, individual citizens and large corporations desperately need encryption to protect their personal privacy and corporate security. On the other hand, this same technology, if used by criminals, can make law enforcements efforts to conduct investigations more difficult. So, we need technology and public policy in response to this dilemma which assure privacy and security, enables law enforcement to continue to do its job, and, at the same time, respects fundamental civil liberties which are at the heart of our constitutional system of government. In my technical judgment, the solution to this problem will not come easily and will certainly not come in one simple step.
The position of the Administration worries many in the technical community since they feel it is but the first step to the banning of any form of encryption except that approved and escrowed by the government. The Clinton Administration says "not on our watch" and I believe them. However, our grandchildren will not have them on the watch when they grow up. So, we need a lasting and effective solution to this problem, not a quick fix that is technologically unsound and unwise as a matter of public policy.
C. Open Technical Processes Are Best For Solving Hard Problems
Even those of us who disagee about Clipper, agree that we are faced with a hard technical and policy problem. So, the question we should be asking is: how does the communications world solve hard problems and are we on track with Clipper? A long history of setting technical standards in the communications industry has shown that solving hard technical problems requires an open design process. The protocols which are the basis of the Internet were not created by ten people in a "smoke-filled room". They evolved openly and sometimes noisily in a highly stressed and highly productive set of interactions. They took over the world and created the basis for the National Information Infrastructure. An examination of the internet standards activity clearly shows the payoff in openness.
The Internet Engineering Task Force (IETF), an open organization attended by all the players with stakes in internet developments, has often faced the situation where two group develop their ideas in different fashions: One group meets in an open public fashion, with constant interactions with different ideas and different economic imperatives of those participating. Another group goes off and sits in smoke-filled rooms designing alternatives in a closed environment. When both are shown to the IETF as a whole, the open design always wins. It has all the rough spots smoothed out and arguments done prior to disclosure. The closed design gets sunk on having to do the smoothing when it is too late. Multimedia Mail, a major future use of the NII, is a good case in point. There are many others examples.
Clipper is in the same boat. While I recognize that a small part of cryptography will always be classified, most of the development of the proposed escrow system has been taking place in those room ( not smoke-filled any more). This process must be brought out into the sunshine of the technical and policy community. Proposals like Clipper must be evolved, if they are to have any chance of success, with the co-operation and understanding of the industrial and academic community and their enthusiastic cooperation rather than their mistrust. This penchant for openness must not be seen as a power struggle between industry and government, or as an excuse for revisiting a decision that technologists dislike for political reasons. Rather it is a reflection of a deep faith in open design processes and a recognition that closed processes invariably lead to solutions which are too narrow and don't last.
A Strawman Proposal
To make my thinking more concrete I have included a much abbreviated design that is predicated on a desire to properly control and authenticate the use of taps and decryption devices to ensure their use only under proper judicial supervision. It is intended to provide a greatly improved and protected mechanism to ensure that proper judicial control is kept and that auditable records exist. I personally believe there are technical safeguards that can be engineered into a key escrow system that could greatly increase privacy and security for users, while ensuring that legitimate law enforcement access is maintained.
The approach requires a change in the warrant mechanism as well as other changes that may require legislative action. I firmly believe that an approach toward the creation of a freely exportable cryptographic system that allows and perhaps even encourages, but does not demand, escrowed keys could, given the right actions on the part of the congress greatly improve the first amendment rights and the custom of personal privacy that exists in the United States and at the same time provide law enforcement with all the lawful access to crypto protected communications that our constitution allows.
The proposed design allows the user of an escrowed crypto-device (ECD) to choose to have his keys escrowed with a set of escrow approved holding organizations (EAHO). An EAHO must conform to a set of rules and audits that define its proper handling of key security, physical and electronic security, management strength, and the ability to obtain a bond to protect the owners of the ECD from illegal actions on the part of EAHO employees. Thus the credibility of the EAHO is examined by the accrediting agency and the organization which provides bonding.
Given that the component keys (CK) have been transported to the chosen EAHOs, the rest of this note describes the process that would be followed in my proposed approach.
When a law enforcement organization (LEO) wants to decrypt a protected conversation, it go through the normal mechanism to find the ECD units being used. The main difference from the existing process is that the warrant they must obtain in order for the EAHOs to release their component keys is as follows.
The LEO obtains a warrant from a judicial source. This warrant is registered as to the date of issue (could use the Bellcore spin-off company) and contains within the warrant the specifics of ECD that will be monitored as well as the length of time of thevalidity of the authorization. The warrant is electronic in form and is then encrypted and signed by the judicial source. It is sent to the LEO. The LEO then sends the warrant along with its additional information to each of the escrow organizations -- this transmission signed by the LEO. The EAHOs check the validity of the LEO, and decrypt the warrant for validity, date and duration. Given this is OK, it sends its part of the key along with the original warrant back to the LEO encrypted under the LEO's public key and signed by the EAHO.
The LEO now can authenticate the EAHO. When the LEO has all parts of the component keys, it ships the messages obtained from the EAHOs -- namely the key parts and the warrant to the decryption "box" where the validity of the warrant, its time of origination and length is again checked before any decryption taking place. If any of these conditions fail, the request for decryption is rejected.
This alternative to Clipper will protect user privacy, ensure legitimate law enforcement access, and has the important advantage of allowing technology and standards to develop in an open, market-drive process.
I would note, too, that the Clinton Administration has shown some interest in pursuing open solutions such as the one I have outlined here. This sentiment is extremely encouraging to me and I would urge the subcommittee to explore this option further with Administration officials.1
D. What's my bottom line on Clipper
Escrow key systems may offer a middle of the road solution to blend the imperatives of technology with the stated but as yet un-substantiated needs of law enforcement. However, in no case, should an escrow system be considered unless it:
* results in an environment which considerable increases the citizens protection against illegal wiretaps
* provides for enabling laws, which would be necessary to establish the electronic warrant system and auditability, must legislate against the mandatory use of clipper technology
* removes export controls on cryptography so as to establish a viable market for American companies to produce products.
III. Digital Telephony
I will enumerate my objections to the proposal from a technical/economic point of view. I join others in attacking it from a privacy/freedom/necessity standpoint.
In the Clipper controversy, one could find some benefits to the nation in a properly designed system. In the case of the Digital Telephony proposal it is difficult to see where there can be a silver lining.
The Digital Telephony proposal calls for the manufacturers of telephone central office switches, tandom switches and SS7 signaling systems to make major design changes to their software systems. These changes would provide, to law enforcement at remote sites, electronic notification of calls to and from selected numbers/individuals as the individuals wander through the complex communications structure that has been created in the United States.
I, and many others, expect that the law would have to be extended to cover PBXs and even personal computers as they are used more and more as personal communications devices.
As a colleague of mine has often said "What guarantees that the job of law enforcement should be easy?". Yet law enforcement noting the increased complexity of the telecommunications systems and maybe even noticing the ability of all the components of the NII to carry such communications, has demanded the right to peek into our minds and to easily find out whom we call and who calls us.
I have often been credited with using the term "the Full Software Employment Act of 1994" when I talk about the Digital Telephony proposal. The proposers talk about costs in the multi hundreds of millions of dollars to be borne by the Government for the first three years. After that all is left open.
As one who has in his time designed and built complex systems and who understands the structure of the current telecommunications structure my reactions are as follows.
Rational estimates obtained from sources in the industry talk about numbers from $1.5 to $3 billion per year. I consider that low. The complexity of just the Plain Old Telephone System software is enormous. Re-designing large and often the most complex parts of it will not be easy nor inexpensive. One must potentially re-engineer the cellular system with its multiple manufacturers plus the local and toll and tandem switching centers. The fact that they are programmed devices makes it feasible but not cheap. The potential for decreased reliability of the national telephone grid caused by the large scale changes (presently undefined) to the software architecture could cause major dangers to the health and economy of the country. If you watch the bugs (errors) that are distributed in well tested and much similar systems (like DOS or MACOS) you can appreciate the opportunities for chaos -- and it must be done in three years.
One should carefully note that the national communications system is marginally reliable at this time. A National Research Council report on it cautioned that it was poorly equipped to survive in the event of catastrophies. The recent set of fiber cuts and the resultant severe disruption of the nation's business is a portent for the future. To spend money that is in short supply satisfying a poorly articulated and poorly justified "problem" with wire tapping is to place the nation's economic health in danger, for communication is the veins that carry the nation's economic blood -- information.
>From the standpoint of the future evolution of our NII, the Digital Telephony proposal presents a major drag. Whenever a new feature is being considered for implementation and marketing, one very important issue will be how much it will cost to implement it in such a manner as to pass the hurdles of the proposal. That could price many good ideas that would improve the usefulness of our NII off the feasibility horizon. Not only would our citizens not have access to these new and useful services but they would not be implemented in US manufactured systems and thus could make our systems less sellable in competition with those of foreign manufacturers of communications equipment for off shore sales. Note that off shore sales in the developing parts of the globe represent major markets which we could lose.
* The Information Age poses hard problems for privacy protection, security and law enforcement
* Increasing amounts of personally sensitive and proprietary information on global, digital networks creates need for security through encryption.
* Technical knowledge of how to create powerful and practically unbreakable encryption technology is available throughout the world.
* Powerful encryption poses real challenges for law enforcement and national security efforts.
* These challenges will not be solved by just a single chip such as Clipper or even just a single legislative act such as the proposed Digital Telephony bill.
* As a closed, secret, inflexible standard, Clipper will not be widely accepted by the market
* Security systems must be trusted by the broad user community. Clipper will never be so trusted because the design of its internal algorithm is secret
* Open systems are the only lasting solution to hard technical problems
* The history of the computer and communications industry has shown that hard technical problems require open flexible solutions arrived at in an open manner.
* The federal government should lead the way toward an open solution, rather than trying in vain to force a closed solution on the problem.
* Real alternative to publicly revealed Clipper system design exist, which meet law enforcement goals while protecting privacy, civil liberties and preserving technological flexibility.
There should be no doubt that new computer and communciations technologies pose challenges for the law enforcement and national security communities. By the same token, for the National Information Infrastructure to succeed, and for US companies to be able to compete in the world market -- the Global Information Infrastructure -- user security and privacy must be protected through robust, open cryptography standards. The Congress and the Administration should lead the way toward solutions that are open, provide for flexibility in the future, and seek an appropriate balance of individual privacy and the legitimate needs of law enforcement. I thank this subcommittee again for its foresight and guidance in holding this hearing and look forward to working with you in the future.
1 I would like to explicitly support an idea originated by Steve Walker of Trusted Information Systems, who proposes the voluntary establishment of software interface standards to crypto devices and crypto software so that manufacturers around the world, but especially in the United States, can create software systems that can be used with nationally required crypto systems around the world. Such an initiative would greatly enhance the privacy of individuals and the market for such systems. It would also create a major opportunity for American companies to market to the world. Such an initiative would assist in the creation of the GII supported by the Vice President